Melt generation, crystallization, and extraction beneath
segmented oceanic transform faults
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Gregg, P. M. et al. “Melt Generation, Crystallization, and
Extraction Beneath Segmented Oceanic Transform Faults.”
Journal of Geophysical Research 114.B11 (2009). ©2009.
American Geophysical Union
As Published
http://dx.doi.org/10.1029/2008jb006100
Publisher
American Geophysical Union (AGU)
Version
Final published version
Accessed
Wed May 25 22:06:23 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/77612
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
Click
Here
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, B11102, doi:10.1029/2008JB006100, 2009
for
Full
Article
Melt generation, crystallization, and extraction beneath segmented
oceanic transform faults
P. M. Gregg,1,2 M. D. Behn,3 J. Lin,3 and T. L. Grove4
Received 11 September 2008; revised 16 April 2009; accepted 21 July 2009; published 13 November 2009.
[1] We examine mantle melting, fractional crystallization, and melt extraction beneath
fast slipping, segmented oceanic transform fault systems. Three-dimensional mantle flow
and thermal structures are calculated using a temperature-dependent rheology that
incorporates a viscoplastic approximation for brittle deformation in the lithosphere.
Thermal solutions are combined with the near-fractional, polybaric melting model of
Kinzler and Grove (1992a, 1992b, 1993) to determine extents of melting, the shape of the
melting regime, and major element melt composition. We investigate the mantle source
region of intratransform spreading centers (ITSCs) using the melt migration approach of
Sparks and Parmentier (1991) for two end-member pooling models: (1) a wide pooling
region that incorporates all of the melt focused to the ITSC and (2) a narrow pooling
region that assumes melt will not migrate across a transform fault or fracture zone.
Assuming wide melt pooling, our model predictions can explain both the systematic
crustal thickness excesses observed at intermediate and fast slipping transform faults as
well as the deeper and lower extents of melting observed in the vicinity of several
transform systems. Applying these techniques to the Siqueiros transform on the East
Pacific Rise we find that both the viscoplastic rheology and wide melt pooling are required
to explain the observed variations in gravity inferred crustal thickness. Finally, we show
that mantle potential temperature Tp = 1350°C and fractional crystallization at depths of
9–15.5 km fit the majority of the major element geochemical data from the Siqueiros
transform fault system.
Citation: Gregg, P. M., M. D. Behn, J. Lin, and T. L. Grove (2009), Melt generation, crystallization, and extraction beneath
segmented oceanic transform faults, J. Geophys. Res., 114, B11102, doi:10.1029/2008JB006100.
1. Introduction
[2] Transform faults are a first-order feature of the global
mid-ocean ridge system. Transform fault offsets range from
50 km to >350 km in length and greatly influence thermal
structure, mantle flow, melting, and crystallization at midocean ridges [Fox and Gallo, 1984; Parmentier and Forsyth,
1985; Phipps Morgan and Forsyth, 1988]. Investigations of
transform faults suggest that the cold, thick lithosphere at
fault offsets effectively divides mid-ocean ridges into
unique segments, focusing melt toward segment centers
and prohibiting shallow across-transform mantle flow.
Gravity and seismic studies at slow spreading ridges indicate along-segment variations in crustal thickness with
enhanced magma accretion at ridge segment centers and
crustal thinning toward transform and nontransform offsets
1
MIT/WHOI Joint Program, Woods Hole Oceanographic Institution,
Woods Hole, Massachusetts, USA.
2
Now at Department of Geosciences, Oregon State University,
Corvallis, Oregon, USA.
3
Department of Geology and Geophysics, Woods Hole Oceanographic
Institution, Woods Hole, Massachusetts, USA.
4
Department of Earth, Atmospheric, and Planetary Sciences,
Massachusetts Institute of Technology, Cambridge, Massachusetts, USA.
Copyright 2009 by the American Geophysical Union.
0148-0227/09/2008JB006100$09.00
[Kuo and Forsyth, 1988; Lin et al., 1990; Lin and Phipps
Morgan, 1992; Tolstoy et al., 1993; Detrick et al., 1995;
Escartin and Lin, 1995; Canales et al., 2000; Hooft et al.,
2000]. Juxtaposing the cold, thick lithosphere of a transform
fault next to a warm steady state ridge segment also
influences the compositions of the erupted lavas. A geochemical signal in the lavas erupted adjacent to transform
faults has been noted by several investigators [Bender et al.,
1978, 1984; Langmuir and Bender, 1984; Langmuir et al.,
1986; Reynolds et al., 1992; Reynolds and Langmuir, 1997].
This effect has been attributed to the influence of the cooler
mantle surrounding the transform fault system, promoting
lower extents of melting near the segment ends.
[3] Oceanic transform faults along fast spreading ridges
are often characterized by segmentation by intratransform
spreading centers (ITSCs) [Menard, 1967; Menard and
Atwater, 1969; Searle, 1983; Fox and Gallo, 1984]. In the
case of several transform fault systems along the fast
spreading East Pacific Rise (EPR) (e.g., Siqueiros, Quebrada,
Discovery, Gofar, and Garrett transform systems, Figure 1a),
fresh basalts sampled on ITSCs and detailed magnetic data
suggest active accretion [Fornari et al., 1989; Carbotte and
Macdonald, 1992; Hekinian et al., 1992, 1995; Perfit et al.,
1996; Nagle et al., 2007]. Segmentation of oceanic transform faults is generally thought to be the result of transtensional forces imposed on the transform fault zone by plate
B11102
1 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 1. Comparison of bathymetry and crustal thickness predictions for Siqueiros transform fault
along the EPR. (a) EPR segmentation and location of transform faults. Note that all of the transform
faults, except for Clipperton, are segmented by at least one ITSC. (b) Siqueiros transform fault
bathymetry with structural interpretation [Fornari et al., 1989]. Solid black lines indicate the seafloor
fabric, circles show locations of seamounts, and dashed black lines show the locations of the fracture
zones and transform fault. (c) Lateral variations in crustal thickness derived from residual mantle
Bouguer gravity calculations [Gregg et al., 2007]. Excess crust is defined as the deviation for the
reference crustal thickness of 6 km, assuming seawater, crust, and mantle densities of 1030, 2730, and
3300 kg m3, respectively.
motion reorganization [Menard and Fisher, 1958; Menard
and Atwater, 1969]. Specifically, the change in spreading
direction enables mantle upwelling beneath the transform
fault domain and the formation of small ITSCs [Searle,
1983; Fornari et al., 1989; Pockalny, 1997; Pockalny et al.,
1997].
[4] Recent analyses of residual mantle Bouguer gravity
anomalies along several intermediate and fast slipping
transform fault systems indicate that for spreading rates
>50 mm/yr full-rate, the observed residual mantle Bouguer gravity anomalies within the transform fault domain
are more negative than their adjacent ridge segments [Gregg
et al., 2007], indicating a mass deficit within these settings.
If the mass deficit is due to crustal thickness variations,
these findings contradict the classic model for transform
faults as regions of thin crust relative to adjacent ridge
segments. One possible explanation for these observations
is that long-lived ITSCs may focus upwelling mantle and
melt beneath the transform domain, warm the fault system,
and result in excess crustal accretion. As we show below,
the apparent contradiction posed by observations of both
thicker crust and lower extents of melting within fast
slipping transform faults is likely a consequence of the
shape of the melt region that feeds the ITSCs.
[5] In this study, we develop 3-D models of a ridgetransform-ridge system to explore the effect of transform
fault segmentation on mantle flow, melt generation, and
melt extraction. Recent modeling investigations of ridgetransform-ridge systems illustrate the importance of incorporating brittle weakening of the lithosphere into thermal
models of these settings [Behn et al., 2007]. We compare
the results of models with both a constant viscosity rheol-
ogy and a temperature-dependent viscoplastic rheology that
approximates brittle deformation [Chen and Morgan, 1990;
Behn et al., 2007]. Calculated thermal structures are used in
conjunction with the fractional melting model of Kinzler
and Grove [1992a, 1992b, 1993] and fractional crystallization model of Yang et al. [1996] to calculate crustal
thickness and predict major element lava compositions.
First, we investigate a suite of generic 3-D transform fault
models segmented by a single ITSC. We find that crustal
production is enhanced at ITSCs relative to the adjacent
ridge segments because melts are pooled at the transform
from a large region of the mantle. Second, we consider a
case appropriate for the Siqueiros transform fault system on
the EPR (Figures 1b and 1c) and compare our modelpredicted crustal thicknesses and lava compositions to
geophysical and geochemical observations in this region.
2. Model Setup
[6] The 3-D mantle flow field and temperature are
calculated using the finite element modeling software
COMSOL 3.3 Multiphysics, which as been benchmarked
for temperature-dependent flow in geologic systems [van
Keken et al., 2008]. Mantle flow is driven by the divergence
of two surface plates moving apart at a half-spreading rate,
U0 (Figure 2). A stress free boundary is assumed at the base
of the model (100 km depth), allowing for convective flux
from the mantle below. The sides of the models are also
stress free. The temperature at the surface of the model is set
to T0 = 0°C and a mantle potential temperature, Tp, is
defined at the surface and then extrapolated along an adiabat
to the base of the model. The model geometry is varied to
2 of 16
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 2. Three-dimensional setup for modeling mantle
temperature structure beneath a segmented transform fault.
Spreading velocities are imposed at the surface. Other
model parameters include spreading rate, U0; temperatures
at the surface, T0, and at the base of the model box, Tm; the
length of transform, LT; and the length of intratransform
spreading center, LITSC. An example melt region is
illustrated with melt fractions, F. The melt pooling width,
Wpool, is shown between the gray dotted lines.
explore the effect of fault segmentation by imposing a
single ITSC of length, LITSC, which bisects a transform
fault of length LT (Figure 2). Model parameters are given in
Table 1, and a list of variables is given in Table 2.
2.1. Numerical Approach
[7] Conservation of mass, momentum, and energy are
given by:
respectively, where r is the density of the mantle, p is
pressure, u is the velocity field, heff is the effective mantle
viscosity, Cp is the heat capacity, T is temperature, and k is
the thermal conductivity.
[8] Mantle flow and temperature are both largely dependent on the assumed mantle viscosity. To explore the
importance of incorporating a temperature-dependent viscoplastic rheology, we compare isoviscous solutions to those
calculated using a temperature-dependent viscosity that
incorporates a viscoplastic approximation for brittle failure
[Chen and Morgan, 1990; Behn et al., 2007]. The effective
mantle viscosity is defined by:
heff ¼
Tmax
Tp
U0
Wpool
zmax
ð4Þ
where htd is the temperature-dependent viscosity, hbyr is a
brittle strength approximation using Byerlee’s law, and hmax
is the maximum viscosity (1023 Pa s). The temperature
dependence of viscosity is calculated by:
Q 1
1
;
htd ¼ h0 exp
R T Tp
ð5Þ
ð1Þ
@u
r
þ u ru ¼ rp þ heff r2 u;
@t
ð2Þ
S
hbyr ¼ pffiffiffi ;
2e_ II
ð3Þ
2
S ¼ ðpr pw Þ;
5
Description
Smoothing coefficient for hydrothermal
cooling equation
Specific heat capacity, J kg1 °K1
Maximum viscosity, Pa s
Reference viscosity, Pa s
Reference thermal conductivity, W m1 °K1
Latent heat of melting, J kg1
Transform fault length, km
Intratransform spreading center length, km
Nusselt number
Activation energy, J mol1
Reference mantle density, kg m3
Universal gas constant, J mol1 °K1
Temperature at the surface of the thermal
model, °C
Maximum temperature of hydrothermal
fluid circulation, °C
Mantle potential temperature used in
melt models, °C
Imposed spreading half-rate, mm yr1
Horizontal length scale for melt
migration, km
Maximum depth of hydrothermal fluid
circulation, km
ð6Þ
where e_ II is the second invariant of the strain rate tensor, and
the maximum shear stress, S, is defined by:
Table 1. Model Parameters
Cp
hmax
h0
k0
L
LT
LITSC
Nu
Q
r
R
T0
!1
r u ¼ 0;
rCp u rT þ kr T ¼ 0;
A
1
1
1
þ
þ
htd hbyr hmax
where h0 is the reference viscosity (1019 Pa s), Q is the
activation energy, and R is the universal gas constant. The
viscosity associated with brittle failure is approximated by
[Chen and Morgan, 1990]:
2
Parameter
B11102
ð7Þ
Value
0.75
Table 2. Variables
3
1.25 10
1023
1019
3
400 103
100
5 – 25
1–8
250 103
3300
8.3114
0
600
1300 – 1400
50
75 – 200
6
Variable
Description
C0
Ci,m
Di,s
Dtc
Initial oxide concentration
Concentration of oxide, i, in the melt
Bulk partition coefficient of oxide, i, in the initial solid
Change in crustal thickness, km
Second invariant of the strain rate tensor, s1
Melt fraction, %
Melt fraction, cation fraction units
Maximum melt fraction, %
Brittle strength viscosity, Pa s
Temperature-dependent viscosity, Pa s
Effective viscosity, Pa s
Effective thermal conductivity, W m1 °K1
Pressure of melting, kbar
Lithostatic pressure, Pa
Hydrostatic pressure, Pa
Bulk partition coefficient of oxide, i, in the melt
Maximum shear stress, Pa
Temperature, °C
Mantle temperature extrapolated from Tp, °C
Solidus temperature, °C
Velocity field, mm yr1
e_ II
F
Fc
Fmax
hbyr
htd
heff
k
p
pr
pw
Pi,m
S
T
Tm
Ts
u
3 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
where k is the effective thermal conductivity, k0 is the
reference thermal conductivity, A is a smoothing factor,
Tmax is the cutoff maximum temperature (600°C), z is depth,
and zmax is the cutoff maximum depth (6 km).
Figure 3. Comparison of melting models for two mantle
rheologies. (a) Comparison of mantle temperature sampled
beneath the oceanic spreading center for an isoviscous and
temperature-dependent viscoplastic rheology. Calculated
melt fraction for (b) constant viscosity rheology and
(c) viscoplastic model. The effect of latent heat of melting
has been incorporated into all melting models. KG92,
Kinzler and Grove [1992a, 1992b, 1993], the solid line;
RJ81, Reid and Jackson [1981], the dotted line; and K03,
Katz et al. [2003], the dashed line. U0 = 50 mm/yr, Tp =
1350°C, Nu = 4, and latent heat of melting L = 400 J/kg.
where pr is the lithostatic pressure and pw is the hydrostatic
pressure.
[9] The incorporation of the viscoplastic rheology promotes focused mantle upwelling and warmer temperatures
in areas with high strain rates [Behn et al., 2007]. This effect
is most pronounced beneath the ridge axis and transform
fault where the imposed surface boundary conditions generate high strain rates. Therefore, models using the viscoplastic rheology result in a warmer thermal structure beneath
the ridge axis and transform fault domain than those
calculated with a constant viscosity rheology (Figure 3a).
[10] Hydrothermal circulation will promote more efficient
heat loss in the crust and upper mantle [Phipps Morgan et
al., 1987; Phipps Morgan and Chen, 1993; Baker et al.,
1996; Maclennan et al., 2005]. This effect is likely to be
particularly important at oceanic transform faults where
extensive brittle deformation increases permeability and
produces pathways for fluids to reach greater depths. We
assume that hydrothermal circulation is present at temperatures <600°C and depths <6 km [Phipps Morgan et al.,
1987; Phipps Morgan and Chen, 1993]. Where hydrothermal circulation is assumed to occur we increase the thermal
conductivity by an enhancement factor or Nusselt number,
Nu [Phipps Morgan et al., 1987]. Because permeability
most likely decreases with depth due to increased confining
pressure and temperature, the enhancement in conductivity
is assumed to decay exponentially as the depth and temperature approach the cutoff values for fluid circulation:
k ¼ k0 þ k0*ð Nu 1Þ* exp
T
z
A* 1 * exp A* 1 ;
Tmax
zmax
2.2. Melt Generation and Extraction
[11] Melting within the mantle at mid-ocean ridges is
dependent on pressure and temperature as well as the
chemical composition of the mantle [Grove et al., 1992;
Langmuir et al., 1992; Asimow and Langmuir, 2003].
Furthermore, the shape of the melting regime influences
the extent of melting and the resulting melt composition
[Forsyth, 1993]. Geochemical melting models often assume
a simple vertical column for partial melting in mid-ocean
environments, but several authors have shown that melt
may be focused from a 3-D region [e.g., Sparks and
Parmentier, 1991] and models of the influence of melt
compositions on melt regime shape have been developed for
major and trace elements [Plank and Langmuir, 1992;
Kinzler, 1997]. As such, we utilize the fractional melting
model of Kinzler and Grove [1992b, 1992a, 1993], in which
melting is a function of temperature, pressure, and mantle
composition. The benefit of this method is that the model
simultaneously tracks residual mantle and melt compositions, as well as extents of melting. Furthermore, the melting
parameterization of Kinzler and Grove [1992a, 1992b,
1993] utilizes extensive experimental results that are tailored specifically to mid-ocean ridge basalts. We calculate
aggregate melt compositions utilizing a shaped melting
regime. Specifically, the extent of melting and resultant
lava composition are predicted using a three-step process:
(1) melt production and composition are calculated at each
point in the melting region assuming polybaric, nearfractional melting; (2) melt migrates vertically until it
reaches the shallowest point of the melt surface and pools;
and (3) in the shallow mantle lithosphere, mantle cooling
induces fractional crystallization of the pooled melt and lava
composition is predicted.
2.2.1. Step 1: Fractional Melting
[12] The polybaric fractional melting model of Kinzler
and Grove (hereafter ‘‘KG92’’) utilizes a parameterized
solidus derived from experimental data to calculate extents
of melting [Kinzler and Grove, 1992b, 1992a, 1993].
Specifically, the solidus temperature, Ts, is given by:
Ts ¼ 1157 þ 16*ð p 0:001Þ 38:7*ð1 Mg#Þ
181*ðNaK#Þ 0*ðTiO2 Þ;
in the spinel stability field (pressure, p 9 kbars) and by:
Ts ¼ 1242 þ 9*ð p 0:001Þ 120*ð1 Mg#Þ 89*ðNaK#Þ
ð10Þ
6:6*ðTiO2 Þ:
in the plagioclase stability field (p < 9 kbars). The oxides
TiO2, Na2O, CaO, and K2O are estimated with the
nonmodal batch melting equation:
C0
;
Ci;m ¼ Di;s þ F 1 Pi;m
ð8Þ
ð9Þ
ð11Þ
where Ci,m is the concentration of oxide (i) in the melt (m),
C0 is the initial mantle concentration, Di,s is the bulk
4 of 16
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Table 3. Estimated MORB Mantle Source Composition
Mantle Compositions
TiO2
SiO2
Al2O3
FeO
MgO
CaO K2O
Na2O
45.0769 0.1311 4.0926 8.2471 39.0478 3.196 0.006 0.2823
W&H
DMMa
Mantle Modesb
W&H
DMM
aug
opx
oliv
plag
aug
opx
oliv
spin
0.128
0.261
0.546
0.065
0.13
0.28
0.57
0.02
a
Bulk depleted MORB mantle composition of Workman and Hart [2005]
(DMM) renormalized excluding Cr2O3, MnO, NiO, and P2O5.
b
Aug, augite; opx, orthopyroxene; oliv, olivine; plag, plagioclase; spin,
spinel.
partition coefficient of oxide (i) in the initial solid (s), Pi,m is
the bulk partition coefficient of the oxide in the melt, and F
is the melt fraction. The oxide concentrations are calculated
iteratively using a bisection method that systematically
moves the endpoints of an initial ‘‘guess’’ interval closer
together until they converge on a value where the error is
minimized. MgO and FeO are also calculated iteratively by
solving the mass balance equations:
B11102
batic and occurs over a range of pressures [Klein and
Langmuir, 1987]. Therefore, we utilize the polybaric incremental batch accumulated melting approach with incomplete melt withdrawal (model 3 in KG92) to approximate
fractional melting of the mantle. This approach assumes a
constant melt production rate of 1% melt per 1 kbar of
ascent. At each melting increment, 99% of the generated
melt is removed and accumulated elsewhere; the remaining
melt and depleted mantle ascend another 1 kbar where they
are further melted by 1% [Ahern and Turcotte, 1979]. As
melting proceeds, the latent heat effect of the melting
reaction acts to decrease the mantle temperature. We assume
a constant latent heat and that the enthalpy of melting does
not change over small extents of melting. We incorporate
the change in temperature, DT, at each step in the calculation as a function of melt fraction, the latent heat of melting,
L, and heat capacity, Cp, using the following formulation
[Turcotte, 1982]:
DT ¼
FL
:
Cp
ð16Þ
[13] It is widely accepted that the process of melt generation beneath a mid-ocean ridge spreading center is adia-
To determine the temperature structure input into the KG92
melting model, the mantle potential temperature, Tp, at the
base of the model domain is set to a value between 1300°–
1400°C. Note that, Tp is defined as the mantle temperature at
the surface extracted along an adiabatic gradient of 0.5°C/km
to depth.
[14] In the original KG92 model, pressure controls the
cessation of melting and fractional melting is assumed to
stop at 4 kbars at the top of a vertical melt column. In our
implementation, melting begins when the adiabatic temperature profile crosses the solidus and continues until conductive cooling from the surface depresses the geotherm
below the solidus. The shape of the melting regime is also
estimated (see section 2.2.2). We assume that the mantle is
described by a four-phase assemblage (olivine, clinopyroxene, orthopyroxene, and an aluminous phase, either spinel
or plagioclase depending on the pressure, see Tables 3 – 5)
and utilize the depleted MORB mantle (DMM) composition
of Workman and Hart [2005] as the initial major element
mantle composition of the mantle (Table 3).
2.2.2. Step 2: Melt Migration
[15] To calculate the final melt composition and the
resultant crustal production, our model must determine
how melt migrates from the melting region to a nearby
spreading center. Previous 3-D melt migration models have
proposed that melt percolating upward through the hot,
permeable mantle will pool along a low permeability
boundary (e.g., the top of the melting region, the base of
the lithosphere, or the cpx-out reaction zone) and then
Table 4. Plagioclase-Lherzolite Meltinga
Table 5. Spinel-Lherzolite Meltinga
½MgOb ¼ ½MgOm Fc ½MgOr ð1 Fc Þ;
ð12Þ
½FeOb ¼ ½FeOm Fc ½FeOr ð1 Fc Þ;
ð13Þ
where [MgO]b is the MgO of the initial bulk mantle
composition, [MgO]m is the MgO of the melt, and [MgO]r is
the MgO of the residual mantle (where the square brackets
denote cation mole percent). Similarly, [FeO]b is the FeO of
the initial mantle composition, [FeO]m is the FeO of the
melt, and [FeO]r is the FeO of the residual mantle. In this
formulation, Fc refers to the fraction of melt produced in
cation fraction units (see Kinzler and Grove [1993] for further
discussion). Once the oxides concentrations are determined
using equations (11) – (13), the Mg #m and NaK #m, the Mg
and NaK numbers of the melt, are calculated as:
½MgOm
½MgOm þ½FeOm
ð14Þ
ðNa2 O þ K2 OÞ
ðNa2 O þ K2 O þ CaOÞ
ð15Þ
Mg#m ¼
NaK#m ¼
Distribution Coefficient
Distribution Coefficientb
b
Oxide
Olivine
Orthopyroxene
Augite
Plagioclase
Oxide
Olivine
Orthopyroxene
Augite
Spinel
K2O
Na2O
TiO2
CaO
0.001
0.001
0.03
0.03
0.001
0.04
0.2
–
0.001
0.15
0.41
–
0.27
1.0
0.001
1.32
K2O
Na2O
TiO2
CaO
0.001
0.001
0.03
0.03
0.001
0.04
0.2
–
0.001
0.2
0.41
–
0.001
0.001
0.135
0.018
a
a
b
b
Pressures < 9 kbar.
Distribution coefficients from Kinzler and Grove [1992a, Table A1].
5 of 16
Pressures 9 kbar.
Distribution coefficients from Kinzler and Grove [1992a, Table A2].
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 4. Depth to the top of the melting region with chosen pooling areas for the (a) constant viscosity
rheology and (b) viscoplastic rheology. Region W (delineated by red outline) delineates the wide melt
pooling model, which pools the entire area of melts that have the potential to migrate toward the ITSC. In
contrast, region N (outlined in blue) assumes that melts will not migrate across the transform fault to the
ITSC. Black arrows indicate the gradient of the melt surface and the direction of melt migration. Tp =
1350°C and Nu = 4.
migrate laterally, ‘‘uphill,’’ along this boundary to the ridge
axis [Sparks and Parmentier, 1991; Sparks et al., 1993;
Magde and Sparks, 1997; Magde et al., 1997; Kelemen and
Aharonov, 1998]. To simulate this two-step melt migration
process, we assume that melt first migrates vertically to the
top of the melting region and then migrates laterally up the
slope of the top of the melting region until it reaches a local
minimum in the surface, which typically corresponds to a
nearby spreading center. Melt migration patterns for the
constant and viscoplastic mantle rheology models are
shown for a 100-km-long transform fault bisected by a
10-km-long ITSC (Figure 4). The depth to the top of the
melt surface is contoured and the gradient of the surface
(black arrows) is plotted to indicate the slope of the melt
surface and the direction the melt will migrate.
[16] Uncertainties with the melt migration model include
(1) how far melt will migrate along this boundary before it
freezes and (2) whether melt will be extracted vertically
before it reaches the ridge axis or ITSC. To explore these
processes, we investigate two end-member scenarios for
melt migration beneath the transform fault domain. The first
scenario assumes a wide pooling region in which all of the
melt projected to migrate to the transform domain will
aggregate at the ITSC (solid red lines in Figure 4). The
capture area delineated by the solid red lines in Figure 4 is
determined using the slope of the melt surface and calculating where the melt will migrate. The second end-member
scenario assumes that melt is unlikely to migrate across a
transform fault or fracture zone due to the increased fracture
permeability within the fault zone; that is, the melt will use
these more-permeable pathways created by the fault and not
continue to the ITSC. In this case, melt that reaches the
ITSC is only pooled from the narrow region of the mantle
delineated between the two fault zones (solid blue lines in
Figure 4). These models will be referred to as migration
model 1, wide pooling region, and model 2, narrow pooling
region, in all subsequent discussions.
[17] Crustal thickness at the ITSC is calculated by integrating the melt production rate over the entire pooling area
that is tapped by the ITSC. The integrated melt production
rate is then divided by the spreading rate and the length of
the ITSC yielding an average crustal thickness for each
ITSC [Forsyth, 1993].
2.2.3. Step 3: Fractional Crystallization
[18] Finally, to approximate fractional crystallization of
the melt prior to eruption we use the approach of Yang et al.
[1996]. This method assumes isobaric fractional crystallization and calculates the fractional crystallization path as a
function of melt composition and pressure between 0.1 to
800 MPa (i.e., 1 bar to 8 kbars). In our models, crystallization pressure is constrained by the depth at which the
melting ceases. Yang et al. [1996] uses algebraic relations
obtained from experimentally determined phase boundaries
for olivine-clinopyroxene-quartz and olivine-clinopyroxeneplagioclase in pseudoternary projections, and constraints
from mineral-melt exchange reactions derived from experimentally produced mineral-melt pairs. This model provides
major element composition for each step of fractional
crystallization.
[19] The composition of the pooled melt is then used to
determine the liquid line of decent (LLD) and to calculate
proxies such as Fe8.0 and Na8.0, which can be compared to
observed major element variations. The LLD represents the
crystallization path that melt will take as it cools in the
shallow parts of the mantle. Transitions in the crystallizing
phase assemblage are illustrated by the kinks in the slopes
of the lines, which in most cases represent the transitions
from the equilibrium phases of olivine + melt, to olivine +
plagioclase + melt, to olivine + plagioclase + clinopyroxene +
melt. Fractionation corrected values of the oxides FeO and
Na2O are used as proxies for depth and extent of melting,
respectively.
3. Numerical Modeling Results
3.1. Benchmark: Comparison to EPR 9°N
[20] To calibrate parameters such as Tp and the maximum
horizontal distance for melt migration, Wpool, lava compositions are first calculated in 2-D and compared to observations of crustal thickness [Canales et al., 2003] and lava
composition [Batiza and Niu, 1992] at 9°N on the EPR. For
a constant viscosity rheology, the crustal thickness calcu-
6 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 5. Comparison of the predicted crustal thickness at the ridge axis for various widths of the melt
pooling regions (Wpool = 200, 150, 100, and 75 km) and hydrothermal cooling (Nu = 1, 4, and 8) for the
(a) constant viscosity rheology and (b) viscoplastic rheology. The light gray shaded region indicates the
seismic crustal thickness variations observed at the 9°N segment of the EPR [Canales et al., 2003].
lated for Tp = 1325°C– 1350°C correlates well with the
seismic observations (Figure 5a). While pooling melt over a
wider region results in greater crustal thickness, incorporating hydrothermal cooling (Nu = 4 and 8) decreases the
predicted crustal thickness (Figure 5a). As discussed above,
the viscoplastic rheology produces a warmer temperature
structure beneath the ridge axis, resulting in thicker crust.
We find that Tp = 1325°C best fits the seismic observations
at EPR 9°N (Figure 5b). Incorporation of hydrothermal
cooling with the viscoplastic rheology further focuses
mantle upwelling beneath the ridge creating larger upwelling velocities and slightly thicker crust (Figure 5b).
[21] We also investigated the sensitivity of Fe8.0 and Na8.0
to Tp and Wpool (Figure 6). Fe8.0 and Na8.0 are normalizations of the wt % FeO and wt % Na2O to 8 wt % MgO
[Klein and Langmuir, 1987]. Fe8.0, which is classically
viewed as a proxy for the mean pressure of melting,
increases as the pressure/depth of melting increases. Since
Na is highly incompatible, Na8.0 is generally used as a
proxy for the extent of melting, decreasing as the extent of
melting increases. As Tp increases the onset of melting
occurs at greater depths, resulting in higher values of Fe8.0
(Figure 6). Additionally, raising Tp increases the extent of
melting and thus Na8.0 is predicted to decrease (Figure 6).
Figure 6. Calculated Na8.0 and Fe8.0 as a function of mantle potential temperature, Tp. Nu = 4 for
models of Wpool = 75 km, 100 km, and 150 km. (a) Constant viscosity and (b) viscoplastic rheology.
Open circles indicate values from individual model runs.
7 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 7. Calculated (a– c) FeO and (d – f) Na2O versus MgO for 2-D models of a ridge with a half
spreading rate of 50 mm/yr, Wpool = 75 km, and Nu = 4. Data points are from major element microprobe
analyses of basalt glass samples collected along the 9°N segment of the EPR by Batiza and Niu [1992].
Crystallizing phases that correspond to changes in the LLD slope are noted in Figures 7a and 7d: ol,
Olivine; pl, Plagioclase; aug, Augite. Figures 7a and 7d show the constant viscosity model with Tp =
1350°C. The solid black line indicates fractional crystallization at top of the melt column and the dashed
line indicates crystallization at 0.2 kbar. Figures 7b and 7e show models for viscoplastic rheology with
Tp = 1325°C. Figures 7c and 7f show models for viscoplastic rheology with Tp = 1350°C.
However, while Fe8.0 and Na8.0 are strongly dependent on
Tp, they vary only slightly with mantle rheology and Wpool.
[22] To calibrate the values for Tp and Wpool used in our
models, we compare our 2-D ridge results to the geochemical observations of Batiza and Niu [1992]. These data
include major element compositions of basaltic glasses from
a comprehensive dredging program at 9– 10°N along the
EPR and its flanks. We find that, Tp = 1350°C, Nu = 4, and
Wpool = 75km results in a reasonably good fit to the
observed FeO and Na2O data (Figures 7a, 7c, 7d, and 7f).
The pressure at the top of the melting triangle varies for
each model run depending on Tp (solid black line in Figure 7).
For Tp = 1350°C the top of the melting region is at 2.4 and
2.0 kbars for the constant viscosity model and the viscoplastic rheology models, respectively.
[23] For the viscoplastic rheology, Tp = 1325°C provides
the best match to the seismically observed crustal thicknesses at the EPR 9°N (Figure 5). However, Tp = 1350°C
better predicts the Na2O and FeO values observed in
microprobe analyses of glass samples from the EPR 9°N.
There is also a discrepancy in the prediction of Fe8.0 and
Na8.0. The Batiza and Niu [1992] data indicate an Fe8.0 9
and Na8.0 2.4, which corresponds to our model predictions for Tp = 1325°C and 1360°C, respectively, assuming
Wpool = 75 km (Figure 6). Based on our 3-D model results,
described later, we believe that these discrepancies arise
from complexities in the pattern of melt migration, which
controls crustal thickness variations, though we cannot rule
out mantle heterogeneity. Because Tp = 1350°C provides the
best match to the major element compositions and Tp =
1325°C best matches the crustal thickness data, we investigate 3-D models in later sections with both of these
temperature constraints.
[24] We also compare the melt productivity predicted by
KG92 with two other melt models, Reid and Jackson [1981]
(hereinafter referred to as RJ81) and Katz et al. [2003]
(hereinafter referred to as K03). RJ81 incorporates a pressure-temperature solidus relationship and calculates melting
linearly as a function of the difference between the temperature of the mantle and of the solidus at a given pressure.
K03 incorporates the Hirschmann [2000] solidus into a
parameterized equation for mantle melting. KG92 differs
from the melting approximations of the RJ81 and K03
models because its solidus is reached at shallower depths
in the mantle [Reid and Jackson, 1981; Katz et al., 2003].
As a result, the KG92 model produces lower extents of
melting than do the K03 and RJ81 models, with mean
extents of melting ranging from 5 to 6% for KG92 to 9 –
10% and 7 –9% for RJ81 and K03, respectively. These
differences become important when estimating production
of crust and the predicted lava composition. To match the
crustal thicknesses observed at the EPR 9°N segment, the
RJ81 and K03 models require much shorter distances of
melt migration to the ridge axis and/or lower values of Tp
compared to KG92.
3.2. Results of 3-D Segmented Transform Fault Results
[25] We investigate the effects of fault segmentation on
melt production and migration by exploring a series of
models with different ITSC lengths (Figure 2). Using the
parameters from the 2-D models that best fit the EPR 9°N
data, we assume Tp = 1350°C, Nu = 4, and U0 of 50 mm/yr
for the constant viscosity models and Tp = 1325°C and
1350°C for the viscoplastic rheology models. Similar to the
2-D results, for a specific value of Tp the principal factor
controlling crustal thickness is the volume of mantle over
8 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
[27] In contrast, cases with a viscoplastic rheology predict
relatively uniform lava composition within the transform
fault domain (Figure 9b). This occurs because the viscoplastic rheology produces an overall warmer thermal structure at the ITSC, which is relatively insensitive to the length
of the ITSC. This indicates that the compositional signature
of the transform fault is sensitive to mantle rheology.
Specifically, models with a constant viscosity predict a
greater transform fault effect on ITSC lava chemistry.
4. Applications to the Siqueiros Transform Fault
System
Figure 8. Calculated change in crustal thickness, Dtc, as a
function of LITSC, where Dtc is defined as the difference
between the predicted 3-D crustal thickness at the ITSC and
a reference model assuming 6 km of crust. Calculations
were done for the (a) constant viscosity and (b) viscoplastic
rheology. Model W incorporates melt from the wide pooling
region (circle symbols), while model N is for the narrow
melt pooling region model (square symbols). Each symbol
is the result of a single 3-D model run. Top and bottom error
bars indicate the range in the calculated crustal thickness by
using the melt surface contour of 0.05 and 0.1, respectively.
Calculations with Tp = 1350°C and Nu = 4 are indicated by
filled symbols, whereas models with Tp = 1325°C and Nu = 4
are indicated by open symbols.
which melt is pooled (Figure 8). This is illustrated by
investigating the change in crustal thickness, Dtc, at the
ITSC as compared to a reference crustal thickness of 6 km.
For the wide melt pooling (model 1) we find that as LITSC
increases, Dtc decreases (Figure 8). This is because, for
LITSC < 25 km, ITSCs pool melt from a similar sized mantle
region. Hence, when the pooled melt is averaged over the
length of the ITSC, there is a larger effect on predicted
crustal thickness for shorter ITSCs. As the length of the
ITSC increases, it behaves more like a steady state 2-D
ridge segment. For narrow pooling (model 2) as the ITSC
length is increased, the predicted crustal thickness increases
(Figure 8). Compared to calculations with a constant viscosity, the viscoplastic rheology model promotes focused
upwelling in regions of increased strain rate beneath the
transform and thus systematically predicts greater variations
in crustal accretion relative to the adjacent ridge segments.
[26] Fault segmentation also influences melt composition
at the ITSCs. For both wide and narrow melt pooling
models, calculations with a constant viscosity predict both
Fe8.0 and Na8.0 to decrease as the length of the ITSC
increases (Figure 9a). These trends suggest that the extent
of melting increases and the average pressure of melting
decreases as the LITSC increases. This is consistent with the
predicted mantle thermal structure, because increasing the
length of the ITCS will eventually create a warmer ridge
segment with higher extents of melting and melting to
shallower depths.
4.1. Tectonic Setting
[28] The Siqueiros transform fault (Figure 1b) has been
the site of several investigations including rock dredging
and Alvin dives [Perfit et al., 1996], magnetic [Carbotte
and Macdonald, 1992] and gravity [Gregg et al., 2007]
studies, and detailed geologic mapping [Fornari et al.,
1989]. This wealth of data makes it an ideal location to
investigate the effects of fault segmentation on melting
and melt extraction. The Siqueiros transform system is a
140-km-long, left-stepping offset located between
8° 20N and 8° 30N on the EPR. The half slip rate of the
transform system is 55 mm/yr [DeMets et al., 1990,
1994]. Bathymetry and side-scan sonar collected from
Siqueiros reveal that the transform system is segmented
into five fault strands separated by four ITSCs labeled A– D
from west to east (Figure 10) [Fornari et al., 1989]. The
current, segmented geometry of Siqueiros transform is
thought to be the result of several discrete periods of
counterclockwise rotation in the spreading direction (4 –
8°) over the past 3 Myr [Pockalny et al., 1997].
[29] Residual mantle Bouguer gravity anomalies calculated along the Siqueiros transform fault system suggest a
mass deficit within the transform fault domain [Gregg et al.,
2007]. Assuming that the entire gravity anomaly is due to
crustal thickness variations, the gravity data indicate excess
crustal thickness (as compared to a reference crust of 6 km)
of up to 1.5 km associated with the bathymetric high of
ITSC D (Figure 2c). Excess crust is also associated with
ITSCs B and C along the central portion of the transform
fault as well as in the inactive fracture zones on either side
of the Siqueiros transform system [Gregg et al., 2007].
Gravity analysis did not reveal excess crust at ITSC A.
[30] We incorporate the geometry of Siqueiros transform
system into a 3-D model to evaluate melting and melt
migration beneath the fault zone. ITSCs A– D range in
length from 10 km at ITSC A to <5 km at ITSC D. Due to
the obliquity of the ITSCs within the transform fault and the
computational difficulty of the exact geometry, the orientations of the ITSCs have been modified to be orthogonal to
the transform fault and parallel to the adjacent ridge segments (Figure 11).
4.2. Crustal Thickness Predictions at Siqueiros
[31] We find that model-predicted crustal thickness variations at the Siqueiros fault system are strongly dependent
on mantle rheology (Figure 11). The constant viscosity
rheology predicts a cooler mantle beneath the transform
system and, subsequently, lower extents of melting compared to the viscoplastic rheology. Along the western
9 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 9. Calculated Fe8.0 and Na8.0 as a function of LITSC for the (a) constant viscosity rheology and
(b) viscoplastic rheology. Model W is for wide melt pooling region, while model N is for narrow melt
pooling region. Gray circles on solid lines indicate calculations for fractional crystallization at 3.0 kbar.
Open circles on dotted lines indicate calculations for fractional crystallization at 0.2 kbar. Each symbol
represents the result of a single model run. Tp = 1350°C and Nu = 4.
portion of Siqueiros, the predicted crustal thickness variations from the wide melt pooling model (model 1) with a
constant viscosity mantle match the gravity-derived crustal
thicknesses calculated for ITSCs A and B (Figure 11c).
However, the predicted crustal thicknesses diminish to the
east and are a poor fit to the gravity-derived crustal thicknesses at ITSCs C and D. In contrast, a wide pooling model
with viscoplastic rheology provides a good fit to the gravityderived crustal thickness data at all four ITSCs (Figure 11d).
[32] The key difference between the viscoplastic and
constant viscosity models is that the viscoplastic rheology
predicts a warmer shallow mantle directly beneath the
transform fault domain and deeper melting near the transform, resulting in higher overall extents of melting. The
warmer temperature structure causes each ITSC to produce
a local minimum in the melt surface thereby focusing melt
extraction and enhancing crustal accretion (Figure 11b).
Finally, the crustal thickness values predicted using narrow
melt pooling (model 2) for both the constant and viscoplastic rheologies are significantly smaller than those derived from residual mantle Bouguer gravity data, indicating
that a wide melt pooling region is necessary to produce the
crustal thickness variations estimated from the gravity data.
4.3. Geochemical Predictions at Siqueiros
[33] We next compare the model-predicted liquid lines of
descent (LLD) to major element analyses of the glassy chill
margins from fresh basalts collected via dredge and Alvin
dive along the Siqueiros fault system (Figure 10) [Perfit et
al., 1996; Hays, 2004]. The maximum pressure at which
fractional crystallization can begin corresponds to the depth
where melting ceases (solid lines in Figures 12 and 13). We
also show a fractional crystallization model for 0.2 kbars,
corresponding to the average seafloor depth (dashed lines in
Figures 12 and 13).
[34] The constant and viscoplastic rheology models compare favorably to the observed FeO composition and there is
very little difference between the wide and narrow pooling
models. The warmer temperatures resulting from the viscoplastic rheology model produce slightly lower FeO implying lower average pressures of melting compared to the
constant viscosity rheology (Figures 12a – 12d). However,
for all models, fractional crystallization beginning at the
depth of melt termination (solid LLD in Figures 12 and 13)
fits the data better than fractional crystallization at shallower
depths (dashed LLD in Figures 12 and 13). The kinks in the
slope of the model calculated LLD change at a different
values of MgO than is seen in the data, which may indicate
that crystallization is polybaric (Figures 12c and 12d). As
such, incorporating polybaric fractional crystallization in
future models may help to better match the trend observed
in the FeO data.
[35] Na2O data from the Siqueiros ITSCs and transform
segments are highly variable spanning a range of 2.2–
2.9 wt % (Figure 13). While the viscoplastic rheology with
wide melt pooling predicts compositions that are in the
middle of the observed range, a single model does not fully
predict the full range of values (Figures 13c and 13d).
Figure 10. Shaded bathymetry for the Siqueiros Transform showing the locations of rock dredge and Alvin dive
samples (colored symbols) collected along the transform
fault [Hays, 2004]. The solid black line indicates the plate
boundary, while the letters denote individual ITSCs.
10 of 16
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
B11102
Figure 11. Results from the Siqueiros Transform melting models. (a) Depth to the top of melting
calculated for a ridge-transform-ridge system with the geometry of Siqueiros and a constant viscosity.
Arrows indicate the local spatial derivative of the melt surface that is assumed to be the melt migration
direction. Red solid lines indicate wide melt pooling regions (W). Blue solid lines indicate the narrow
melt pooling regions (N), which assumes that melt does not migrate across the transform fault. (b) Depth
to the top of the melt surface for the viscoplastic rheology case. (c) Calculated and estimated crustal
thickness values for the constant viscosity model. Tp = 1350°C and Nu = 4. Gray bars show the gravityderived crustal thicknesses at the Siqueiros transform fault from Gregg et al. [2007]. Red bars indicate the
predicted crustal thickness for the wide pooling region model. Blue bars indicate the predicted crustal
thicknesses for the narrow pooling region model. Note that no crustal accretion is predicted at spreading
center D, as it is not a local minimum in the melt surface. (d) Crustal thickness predictions for the
viscoplastic rheology. Tp = 1325°C and Nu = 4. Note that this model does a much better job of predicting
the crustal thicknesses as observed from the gravity calculations. Crustal thickness predictions for Tp =
1350°C are indicated by the dotted lines.
Pooling melt from a narrow region of the mantle within the
transform fault domain may explain some of the lower
values of Na2O values seen in particular at ITSC A, which
is regarded as a long-lived ITSC [Fornari et al., 1989]. On
the other hand, utilizing a lower Tp of 1325°C results in
melting at shallower depths and lower extents of melting,
which may explain the higher Na2O values observed at
ITSC C and the transform segment between ITSCs B and C.
The constant viscosity model underpredicts the extent of
melting, i.e., Na2O values that are too high, for both wide
and narrow pooling (Figures 13a and 13b). This is consistent with the underprediction of crustal thickness using the
constant viscosity model. In summary, the viscoplastic
rheology with wide pooling and Tp = 1350°C does the best
job of fitting both the gravity-inferred crustal thickness data
as well as the majority of the geochemical data (Figures 12
and 13).
5. Discussion
5.1. Factors Controlling Crustal Thickness Variations
[36] In General, melt migration beneath mid-ocean ridge
spreading centers is poorly constrained. In 2-D ridge models, the predicted crustal thickness variations are primarily
sensitive to Wpool, Tp, and the assumed melting model. For
example, for U0 = 50 mm/yr, Wpool = 75 km, and Tp =
1350°C, the KG92 melt model predicts a crustal thickness
of 6 – 7 km, while the K03 and RJ81 melting models predict
thicker crust on the order of 8 – 10 km. Therefore, to match
the crustal thickness variations observed at the EPR 9°N
segment, the RJ81 and K03 models require the migration of
melt over shorter distances to the ridge axis or lower values
for Tp.
[37] A similar relationship is observed in 3-D at ITSCs
within a transform fault system. Previous studies have
assumed that melts extruded at ITSCs came from directly
below the transform fault domain or from the nearby ridge
axis [Fornari et al., 1989; Hekinian et al., 1992, 1995;
Perfit et al., 1996; Wendt et al., 1999], and have not
discussed the possibility that melts may be drawn in across
the transform fault from a large volume of the surrounding
mantle. If we assume that the negative residual mantle
Bouguer gravity anomaly observed along the Siqueiros fault
system is primarily due to crustal thickness variations
[Gregg et al., 2007], our calculations show that melts must
focus into the transform fault from a large volume of the
mantle. The wide pooling model for the Siqueiros transform
predicts that melts travel, on average, 40 km to reach an
11 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 12. Liquid lines of decent (LLD) model calculations of FeO versus MgO. Colored LLD in each
plot correspond to the color of the data point for each ITSC or transform fault strand as shown in the
legend on the bottom right and in Figure 10. Solid LLD lines indicate fractional crystallization at the top
of the melt surface, which ranges in depth from 18 to 24 km (6 –8 kbar) for constant viscosity and 9 –
15.5 km (3– 5 kbar) for viscoplastic rheology, while dashed LLD lines indicate fractional crystallization
at 0.2 kbar, which would be expected for fractional crystallization in a shallow magma lens. All models
were run for Tp = 1350°C and Nu = 4. The gray shaded regions indicate the predicted model region for
Tp = 1325°C. Models were run for (a and b) constant viscosity and (c and d) viscoplastic rheology and for
a wide melt pooling region (Figures 12a and 12c) and a narrow melt pooling region (Figures 12b and 12d).
ITSC. This is approximately half the distance that melts are
required to migrate (75 km) in our 2-D melting models to
match the EPR 9°N data. However, since there are fresh
lava flows observed along the fault strands separating the
ITSCs, it is clear that some of the melt does not make it to
the spreading segments [Perfit et al., 1996]. Overall, we
conclude that to generate enhanced crustal thicknesses
within fast slipping transform systems by deep melt migration, melt must migrate into the transform domain from a
wide region of underlying mantle.
[38] Gravity derived crustal thickness variations at the
Garrett transform fault [Gregg et al., 2007] also suggest that
melt is tapped from a wide region of the mantle. The Garrett
transform is 130 km long with three ITSCs containing
four ridges of 9 km long that lie in between four fault
strands [Hekinian et al., 1992]. Peridotites collected from
the Garrett transform are very depleted, indicating high
degrees of melting in the mantle regions surrounding the
fault system [Niu and Hekinian, 1997a, 1997b; Constantin,
1999]. However, the melt generated in the mantle surrounding the Garrett transform does not appear to migrate directly
into the transform fault and aggregate at the ITSCs. Rather,
the majority of the excess crust observed at the Garrett
system is associated with the high topography flanking the
transform fault and fracture zones and not with the ITSCs,
which have crustal thicknesses equal to or less than the
adjacent ridge segments [Gregg et al., 2007]. These observations suggest that the melts pooled from a wide mantle
region surrounding the Garrett system are extruded on the
flanks of the transform fault where more permeable pathways may have been produced by brittle deformation.
5.2. Effect of Siqueiros Transform on Mantle Melting
[39] The geochemical transform fault effect suggests that
a cooler mantle region surrounding a transform fault system
will promote deeper and lower extents of melting [Bender et
al., 1978, 1984; Langmuir and Bender, 1984; Langmuir et
al., 1986; Reynolds et al., 1992; Reynolds and Langmuir,
1997]. This effect can be observed by comparing the major
element compositions of lavas collected along the EPR 9°N
segment and the Siqueiros transform fault system [Batiza and
Niu, 1992; Perfit et al., 1996; Hays, 2004] using the Fe8.0
and Na8.0 variables. Lavas from the EPR 9°N ridge segment
have lower values of Fe8.0 (9 wt % FeO, Figures 7 and 12),
indicating a shallower mean pressure of polybaric melting. At
the Siqueiros transform fault, erupted lavas have significantly
higher values of Fe8.0 (10 – 11 wt % FeO, Figure 12),
indicating that the mean pressure of melting is deeper. This
trend is also observed in the predicted compositions from
our models. Thus, the transform fault effect noted by
Langmuir and Bender [1984] is present in some major
elements and is recoverable in our models.
[40] As described above, the incorporation of the viscoplastic mantle rheology creates a warmer thermal structure
beneath the transform fault domain and subsequently lower
pressures of melt cessation. This effect, coupled with
12 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
B11102
Figure 13. Liquid lines of decent (LLD) model calculations of Na2O versus MgO. Colored LLD in each
plot correspond to color of the data point for each ITSC or transform fault strand as shown in the legend
on the bottom right and in Figure 10. Solid LLD lines indicate fractional crystallization at the top of the
melt surface, which ranges in depth from 18 to 24 km (6 – 8 kbar) for constant viscosity and 9 –15.5 km
(3 – 5 kbar) for viscoplastic rheology, while dashed LLD lines indicate fractional crystallization at
0.2 kbar, which would be expected for fractional crystallization in a shallow magma lens. All models
were run for Tp = 1350°C and Nu = 4. The gray shaded regions indicate the model region for Tp =
1325°C. Models were run for (a and b) constant viscosity and (c and d) viscoplastic rheology for a wide
melt pooling region (Figures 13a and 13c) and a narrow melt pooling region (Figures 13b and 13d). The
two gray shaded regions in Figure 13d indicate the regions obtained for Tp = 1325°C and arise from the
two clusters observed in the LLDs.
focusing melt from a large volume of the mantle around the
transform fault, results in predicted crustal thicknesses at the
ITSCs that are comparable to those observed at a 2-D ridge
segment. Therefore, not only does the viscoplastic rheology
model match the geophysical and geochemical observations
at Siqueiros, this rheology does a very good job of explaining why the ITSCs of Siqueiros are producing volumes of
lava equivalent to those produced at the EPR while simultaneously predicting deeper, lower extents of melting.
5.3. High-Pressure Fractional Crystallization
[41] While high-pressure fractional crystallization has
been proposed in previous studies [Michael and Cornell,
1998; Herzberg, 2004; Eason and Sinton, 2006] to explain
compositional variability at ridges with low magma flux,
Langmuir et al. [1992] maintain that there are many problems associated with this mechanism. Langmuir et al.
[1992] assert that in a ridge setting the lithosphere must
be sufficiently thick to shut melting off at depth and for
high-pressure crystallization to be predicted it must be
consistent with melting processes. By combining fractional
melting and crystallization models with mantle thermal
models we are able to quantify both the termination of
melting and the onset of crystallization and we illustrate that
high-pressure fractional crystallization may occur in regions
with thicker lithosphere such as beneath transform faults. At
the 9°N segment of the EPR our models predict fractional
crystallization depths of 0 – 3 km (0.2 – 1.0 kbar). These
predictions are consistent with previous estimations for
shallow crystallization beneath the EPR ridge axis [e.g.,
Michael and Cornell, 1998], while matching observed
major element compositions [Batiza and Niu, 1992] and
crustal thickness variations [Canales et al., 2003]. On the
other hand, we find that models with fractional crystallization at greater depths, 9 – 18 km (3 – 6 kbars), fit the
observed variations in composition at the Siqueiros transform fault ITSCs [Perfit et al., 1996; Hays, 2004]. In fact,
models of shallow crystallization do a very poor job of
predicting the observed major element variations at the
Siqueiros transform fault. Specifically, if shallow level
fractional crystallization were occurring we would expect
the observed compositions to fall upon the 0.2 kbar line in
Figures 12 and 13. Since our models take into account the
full melting process and begin crystallization at depths
above melt termination, they are self-consistent given the
physical constraints of melting and crystallization processes.
5.4. Melt Extraction at Siqueiros Transform Fault
[42] The observed major and trace element data from the
Siqueiros transform system vary widely from enriched midocean ridge basalt (EMORB) to depleted mid-ocean ridge
basalt (DMORB). However, lavas from a single tectonic
region within Siqueiros are relatively homogenous (e.g.,
Figure 13) [Perfit et al., 1996; Hays et al., 2004]. These
13 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
findings have led previous researchers to suggest that there
are multiple mantle sources beneath the transform fault and
that each ITSC and fault strand is independent of its
neighbor and taps a different melt lens and/or mantle supply
[Fornari et al., 1989; Perfit et al., 1996; Hays, 2004]. These
models implicitly assume that the melt feeding into the
transform domain must either come from directly below the
transform fault or from the nearby ridge axis. In contrast,
our 3-D thermal and melt models for Siqueiros indicate that
the mantle flanking the transform fault supplies melt to the
ITSCs. Specifically, by inferring the volume of the melting
regime using the melt migration techniques of Sparks and
Parmentier [1991] and then fractionally crystallizing the
pooled melts at the depth of melt termination, we can reproduce the observed range in FeO and Na2O (Figures 12c,
12d, 13c, and 13d).
[43] The higher Fe8.0 values observed in the melts erupted
at the ITSCs are likely a consequence of the large volume of
deep mantle melts which are generated in the mantle
adjacent to the transform fault and contribute to the pooled
melts feeding the ITSCs. High-pressure fractional crystallization beneath the transform fault leads to earlier enrichment in FeO, which will further contribute to high Fe8.0
values. The results of our models indicate that the majority
of melts extracted at Siqueiros are aggregated from a large
volume of mantle weighted toward deeper low-extent melts
and the pooled melt collected from this volume experiences
fractional crystallization at depths ranging from 9 –18 km.
This suggests that the variations in the composition of lavas
from Siqueiros transform fault can be explained by the
complexities of migrating melt into the transform fault, and
the thermal effect of cold lithosphere in the fracture zone.
6. Conclusions
[44] Our investigation of melting, fractional crystallization, and melt extraction at segmented oceanic transform
faults illustrates that, to explain the gravity-derived crustal
thickness data from fast and intermediate slipping transforms, melt migration from a wide region of the mantle
surrounding the transform fault is required. Significantly,
these pooled melts contain a higher proportion of deep nearfractional melts that collect and erupt in the transform fault
environment. We find that, depending on the length of an
ITSC, crustal thickness at ITSCs is enhanced by 2 –3 km for
models incorporating an isoviscous rheology and up to 3 –
6.5 km for a temperature-dependent viscoplastic rheology.
To match the gravity-derived crustal thickness variations
observed within the Siqueiros transform fault (>1.5 km),
melt must aggregate from a wide pooling area that extends
up to 40 km on either side of the transform.
[45] Our model produces a geochemical transform fault
effect of low extents of melting by pooling melts from a
large volume of the mantle around the transform where
melting shuts off at greater depths compared to beneath the
adjacent ridge axes. This leads to increased crustal production in the transform while simultaneously producing
pooled aggregate near-fractional melt compositions that
are dominated by low-extent, deep melts. Finally, utilizing
a temperature-dependent viscoplastic rheology, with Tp =
1350°C, in conjunction with the fractional melting model of
Kinzler and Grove [1992a, 1992b, 1993] and the fractional
B11102
crystallization model of Yang et al. [1996], we demonstrate
that the observed variations in major element composition
from the Siqueiros transform fault can be explained by
pooling melts from a wide region of the mantle and fractionally crystallizing them at the depth of melt termination.
[46] Acknowledgments. We are grateful for helpful discussions with
H. Dick, D. Forsyth, M. Jackson, M. Krawczynski, S. Hart, E. Roland,
K. Sims, H. Schouten, Z. Wang, C. Waters, and the WHOI Geochemistry
and Geophysics groups. This manuscript was improved by insightful
reviews by J. Sinton and T. Sisson. We thank H. Yang and R. Kinzler for
use of their original FORTRAN codes for developing our models. We also
thank M. Perfit and M. Hays for use of their Siqueiros geochemical
database. This research was supported by WHOI Academic Programs
Office (PMG), NSF grants OCE-0649103 and OCE-0623188 (MDB),
and the Charles D. Hollister Endowed Fund for Support of Innovative
Research at WHOI (J.L.).
References
Ahern, J. L., and D. L. Turcotte (1979), Magma migration beneath and
ocean ridge, Earth Planet. Sci. Lett., 45, 115 – 122, doi:10.1016/0012821X(79)90113-4.
Asimow, P. D., and C. H. Langmuir (2003), The importance of water to
oceanic mantle melting regimes, Nature, 421, 815 – 820, doi:10.1038/
nature01429.
Baker, E. T., Y. J. Chen, and J. Phipps Morgan (1996), The relationship
between near-axis hydrothermal cooling and the spreading rate of midocean ridges, Earth Planet. Sci. Lett., 142, 137 – 145, doi:10.1016/0012821X(96)00097-0.
Batiza, R., and Y. L. Niu (1992), Petrology and magma chamber processes
at the East Pacific Rise similar to 9°300N, J. Geophys. Res., 97, 6779 –
6797, doi:10.1029/92JB00172.
Behn, M. D., M. S. Boettcher, and G. Hirth (2007), Thermal structure of
oceanic transform faults, Geology, 35, 307 – 310, doi:10.1130/
G23112A.1.
Bender, J. F., F. N. Hodges, and A. E. Bence (1978), Petrogenesis of basalts
from project Famous Area—Experimental-study from 0-kbar to 15-kbars,
Earth Planet. Sci. Lett., 41, 277 – 302, doi:10.1016/0012-821X(78)
90184-X.
Bender, J. F., C. H. Langmuir, and G. N. Hanson (1984), Petrogenesis of
basalt glasses from the Tamayo region, East Pacific Rise, J. Petrol., 25,
213 – 254.
Canales, J. P., R. S. Detrick, J. Lin, J. A. Collins, and D. R. Toomey (2000),
Crustal and upper mantle seismic structure beneath the rift mountains
and across a nontransform offset at the Mid-Atlantic Ridge (35°N),
J. Geophys. Res., 105, 2699 – 2719, doi:10.1029/1999JB900379.
Canales, J. P., R. Detrick, D. R. Toomey, and S. D. Wilcock (2003),
Segment-scale variations in the crustal structure of 150 – 300 kyr old fast
spreading oceanic crust (East Pacific Rise, 8°150N – 10°50N) from wideangle seismic refraction profiles, Geophys. J. Int., 152, 766 – 794,
doi:10.1046/j.1365-246X.2003.01885.x.
Carbotte, S., and K. Macdonald (1992), East Pacific Rise 8° – 10°300N—
Evolution of ridge segments and discontinuities from Seamarc II and
3-dimensional magnetic studies, J. Geophys. Res., 97, 6959 – 6982,
doi:10.1029/91JB03065.
Chen, Y. J., and W. J. Morgan (1990), A nonlinear rheology model for midocean ridge axis topography, J. Geophys. Res., 95, 17,583 – 17,604,
doi:10.1029/JB095iB11p17583.
Constantin, M. (1999), Gabbroic intrusions and magmatic metasomatism in
harzburgites from the Garrett transform fault: Implications for the nature
of the mantle-crust transition at fast-spreading ridges, Contrib. Mineral.
Petrol., 136, 111 – 130, doi:10.1007/s004100050527.
DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein (1990), Current plate
motions, Geophys. J. Int., 101, 425 – 478, doi:10.1111/j.1365-246X.1990.
tb06579.x.
DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein (1994), Effect of
recent revisions to the geomagnetic reversal time-scale on estimates of
current plate motions, Geophys. Res. Lett., 21, 2191 – 2194, doi:10.1029/
94GL02118.
Detrick, R. S., H. D. Needham, and V. Renard (1995), Gravity-anomalies
and crustal thickness variations along the Mid-Atlantic Ridge between
33°N and 40°N, J. Geophys. Res., 100, 3767 – 3787, doi:10.1029/
94JB02649.
Eason, D., and J. Sinton (2006), Origin of high-Al N-MORB by fractional
crystallization in the upper mantle beneath the Galápagos Spreading
Center, Earth Planet. Sci. Lett., 252, 423 – 436, doi:10.1016/
j.epsl.2006.09.048.
14 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
Escartin, J., and J. Lin (1995), Ridge offsets, normal faulting, and gravity
anomalies of slow spreading ridges, J. Geophys. Res., 100, 6163 – 6177,
doi:10.1029/94JB03267.
Fornari, D. J., D. G. Gallo, M. H. Edwards, J. A. Madsen, M. R. Perfit, and
A. N. Shor (1989), Structure and topography of the Siqueiros transformfault system—Evidence for the development of intra-transform spreading
centers, Mar. Geophys. Res., 11, 263 – 299, doi:10.1007/BF00282579.
Forsyth, D. W. (1993), Crustal thickness and the average depth and degree
of melting in fractional melting models of passive flow beneath mid-ocean
ridges, J. Geophys. Res., 98, 16,073 – 16,079, doi:10.1029/93JB01722.
Fox, P. J., and D. G. Gallo (1984), A tectonic model for ridge-transform-ridge
plate boundaries: Implications for the structure of oceanic lithosphere,
Tectonophysics, 104, 205 – 242, doi:10.1016/0040-1951(84)90124-0.
Gregg, P. M., J. Lin, M. D. Behn, and L. G. J. Montesi (2007), Spreading
rate dependence of gravity anomalies along oceanic transform faults,
Nature, 448, 183 – 187, doi:10.1038/nature05962.
Grove, T. L., R. J. Kinzler, and W. B. Bryan (1992), Fractionation of midocean ridge basalt, in Mantle Flow and Melt Generation at Mid-Ocean
Ridges, Geophys. Monogr. Ser., vol. 71, edited by J. Phipps Morgan et al.,
pp. 281 – 310, AGU, Washington, D. C.
Hays, M. R. (2004), Intra-transform volcanism along the Siqueiros Fracture
Zone 8°200N – 8°300N, East Pacific Rise, Ph.D. thesis, 251 pp., Univ. of
Fla., Gainesville.
Hays, M. R., M. R. Perfit, D. J. Fornari, and W. Ridley (2004), Magmatism
in the Siqueiros transform: Major and trace element evidence for mixing
and multiple sources, Eos Trans. AGU, 85(47), Fall Meet. Suppl.,
Abstract T41A-1165.
Hekinian, R., D. Bideau, M. Cannat, J. Francheteau, and R. Hebert
(1992), Volcanic activity and crust mantle exposure in the ultrafast
Garrett transform-fault near 13°280S in the Pacific, Earth Planet. Sci.
Lett., 108, 259 – 275, doi:10.1016/0012-821X(92)90027-S.
Hekinian, R., D. Bideau, R. Hebert, and Y. L. Niu (1995), Magmatism in
the Garrett transform-fault (East Pacific Rise near 13°270S), J. Geophys.
Res., 100, 10,163 – 10,185, doi:10.1029/94JB02125.
Herzberg, C. (2004), Partial crystallization of mid-ocean ridge basalts in the
crust and mantle, J. Petrol., 45, 2389 – 2405, doi:10.1093/petrology/
egh040.
Hirschmann, M. M. (2000), Mantle solidus: Experimental constraints and
the effects of peridotite composition, Geochem. Geophys. Geosyst., 1(10),
1042, doi:10.1029/2000GC000070.
Hooft, E. E. E., R. S. Detrick, D. R. Toomey, J. A. Collins, and J. Lin
(2000), Crustal thickness and structure along three contrasting spreading
segments of the Mid-Atlantic Ridge, 33.5° – 35°N, J. Geophys. Res., 105,
8205 – 8226, doi:10.1029/1999JB900442.
Katz, R. F., M. Spiegelman, and C. H. Langmuir (2003), A new parameterization of hydrous mantle melting, Geochem. Geophys. Geosyst., 4(9),
1073, doi:10.1029/2002GC000433.
Kelemen, P. B., and E. Aharonov (1998), Periodic formation of magma
fractures and generation of layered gabbros in the lower crust beneath
oceanic spreading centers, in Faulting and Magmatism at Mid-Ocean
Ridges, Geophys. Monogr. Ser., vol. 106, edited by W. R. Buck et al.,
348 pp., AGU, Washington, D. C.
Kinzler, R. J. (1997), Melting of mantle peridotite at pressures approaching
the spinel to garnet transition: Application to mid-ocean ridge basalt
petrogenesis, J. Geophys. Res., 102, 853 – 874, doi:10.1029/96JB00988.
Kinzler, R. J., and T. L. Grove (1992a), Primary magmas of mid-ocean
ridge basalts: 2. Applications, J. Geophys. Res., 97, 6907 – 6926,
doi:10.1029/91JB02841.
Kinzler, R. J., and T. L. Grove (1992b), Primary magmas of mid-ocean
ridge basalts: 1. Experiments and methods, J. Geophys. Res., 97, 6885 –
6906, doi:10.1029/91JB02840.
Kinzler, R. J., and T. L. Grove (1993), Corrections and further discussion of
the primary magmas of mid-ocean ridge basalts, 1 and 2, J. Geophys.
Res., 98, 22,339 – 22,347, doi:10.1029/93JB02164.
Klein, E. M., and C. H. Langmuir (1987), Global correlations of ocean
ridge basalt chemistry with axial depth and crustal thickness, J. Geophys.
Res., 92, 8089 – 8115, doi:10.1029/JB092iB08p08089.
Kuo, B. Y., and D. W. Forsyth (1988), Gravity anomalies of the ridge
transform intersection system in the South Atlantic between 31 and
34.5°S: Upwelling centers and variations in crustal thickness, Mar. Geophys. Res., 10, 205 – 232, doi:10.1007/BF00310065.
Langmuir, C. H., and J. F. Bender (1984), The Geochemistry of oceanic
basalts in the vicinity of transform faults—Observations and implications,
Earth Planet. Sci. Lett., 69, 107 – 127, doi:10.1016/0012-821X(84)
90077-3.
Langmuir, C. H., J. F. Bender, and R. Batiza (1986), Petrological and
tectonic segmentation of the East Pacific Rise, 5°300 – 14°300N, Nature,
322, 422 – 429, doi:10.1038/322422a0.
Langmuir, C. H., E. M. Klein, and T. Plank (1992), Petrological systematics
of mid-ocean ridge basalts: Constraints on melt generation beneath ocean
B11102
ridges, in Mantle Flow and Melt Generation at Mid-Ocean Ridges,
Geophys. Monogr. Ser., vol. 71, edited by J. Phipps Morgan et al.,
pp. 183 – 280, AGU, Washington, D. C.
Lin, J., and J. Phipps Morgan (1992), The spreading rate dependence of
three-dimensional mid-ocean ridge gravity structure, Geophys. Res. Lett.,
19, 13 – 16, doi:10.1029/91GL03041.
Lin, J., G. M. Purdy, H. Schouten, J.-C. Sempere, and C. Zervas (1990),
Evidence from gravity data for focused magmatic accretion along the
Mid-Atlantic Ridge, Nature, 344, 627 – 632, doi:10.1038/344627a0.
Maclennan, J., T. Hulme, and S. C. Singh (2005), Cooling of the lower
oceanic crust, Geology, 33, 357 – 360, doi:10.1130/G21207.1.
Magde, L. S., and D. W. Sparks (1997), Three-dimensional mantle upwelling, melt generation, and melt migration beneath segmented slow spreading ridges, J. Geophys. Res., 102, 20,571 – 20,583, doi:10.1029/
97JB01278.
Magde, L. S., D. W. Sparks, and R. S. Detrick (1997), The relationship
between buoyant mantle flow, melt migration, and gravity bull’s eyes at
the Mid-Atlantic Ridge between 33°N and 35°N, Earth Planet. Sci. Lett.,
148, 59 – 67, doi:10.1016/S0012-821X(97)00039-3.
Menard, H. W. (1967), Extension of northeastern-Pacific fracture zones,
Science, 155, 72 – 74, doi:10.1126/science.155.3758.72.
Menard, H. W., and T. Atwater (1969), Origin of fracture zone topography,
Nature, 222, 1037 – 1040, doi:10.1038/2221037a0.
Menard, H. W., and R. L. Fisher (1958), Clipperton fracture zone in the
northeastern equatorial Pacific, J. Geol., 66, 239 – 253, doi:10.1086/
626502.
Michael, P. J., and W. C. Cornell (1998), Influence of spreading rate and
magma supply on crystallization and assimilation beneath mid-ocean
ridges: Evidence from chlorine and major element chemistry of mid-ocean
ridge basalts, J. Geophys. Res., 103, 18,325 – 18,356, doi:10.1029/
98JB00791.
Nagle, A. N., R. C. Pickle, A. E. Saal, E. H. Hauri, and D. W. Forsyth
(2007), Volatiles in basalts from intra-transform spreading centers:
Implications for melt migration models, Eos Trans. AGU, 88(52), Fall
Meet. Suppl., Abstract DI43A-05.
Niu, Y., and R. Hekinian (1997a), Spreading-rate dependence of the extent
of mantle melting beneath ocean ridges, Nature, 385, 326 – 329,
doi:10.1038/385326a0.
Niu, Y. L., and R. Hekinian (1997b), Basaltic liquids and harzburgitic
residues in the Garrett Transform: A case study at fast-spreading ridges,
Earth Planet. Sci. Lett., 146, 243 – 258, doi:10.1016/S0012-821X(96)
00218-X.
Parmentier, E. M., and D. W. Forsyth (1985), Three-dimensional flow
beneath a slow spreading ridge axis: A dynamic contribution to the
deepening of the median valley toward fracture zones, J. Geophys.
Res., 90(B1), 678 – 684, doi:10.1029/JB090iB01p00678.
Perfit, M. R., et al. (1996), Recent volcanism in the Siqueiros transform
fault: Picritic basalts and implications for MORB magma genesis, Earth
Planet. Sci. Lett., 141, 91 – 108, doi:10.1016/0012-821X(96)00052-0.
Phipps Morgan, J., and Y. J. Chen (1993), The genesis of oceanic-crust:
Magma injection, hydrothermal circulation, and crustal flow, J. Geophys.
Res., 98, 6283 – 6297, doi:10.1029/92JB02650.
Phipps Morgan, J., and D. W. Forsyth (1988), Three-dimensional flow and
temperature perturbations due to a transform offset: Effects on oceanic
crustal and upper mantle structure, J. Geophys. Res., 93, 2955 – 2966,
doi:10.1029/JB093iB04p02955.
Phipps Morgan, J., E. M. Parmentier, and J. Lin (1987), Mechanisms for the
origin of mid-ocean ridge axial topography: Implications for the thermal
and mechanical structure of accreting plate boundaries, J. Geophys. Res.,
92, 12,823 – 12,836, doi:10.1029/JB092iB12p12823.
Plank, T., and C. H. Langmuir (1992), Effects of the melting regime on the
composition of the oceanic crust, J. Geophys. Res., 97, 19,749 – 19,770,
doi:10.1029/92JB01769.
Pockalny, R. A. (1997), Evidence of transpression along the Clipperton
Transform: Implications for processes of plate boundary reorganization,
Earth Planet. Sci. Lett., 146, 449 – 464, doi:10.1016/S0012-821X(96)
00253-1.
Pockalny, R. A., P. J. Fox, D. J. Fornari, K. C. Macdonald, and M. R. Perfit
(1997), Tectonic reconstruction of the Clipperton and Siqueiros Fracture
Zones: Evidence and consequences of plate motion change for the last
3 Myr, J. Geophys. Res., 102, 3167 – 3181, doi:10.1029/96JB03391.
Reid, I., and H. R. Jackson (1981), Oceanic spreading rate and crustal
thickness, Mar. Geophys. Res., 5, 165 – 172.
Reynolds, J. R., and C. H. Langmuir (1997), Petrological systematics of the
Mid-Atlantic Ridge south of Kane: Implications for ocean crust formation, J. Geophys. Res., 102, 14,915 – 14,946, doi:10.1029/97JB00391.
Reynolds, J. R., C. H. Langmuir, J. F. Bender, K. A. Kastens, and W. B. F.
Ryan (1992), Spatial and temporal variability in the geochemistry of
basalts from the East Pacific Rise, Nature, 359, 493 – 499, doi:10.1038/
359493a0.
15 of 16
B11102
GREGG ET AL.: MELTING AT OCEANIC TRANSFORM FAULTS
Searle, R. C. (1983), Multiple, closely spaced transform faults in fast-slipping
fracture-zones, Geology, 11, 607 – 610, doi:10.1130/0091-7613(1983)11<
607:MCSTFI>2.0.CO;2.
Sparks, D. W., and E. M. Parmentier (1991), Melt extraction from the
mantle beneath spreading centers, Earth Planet. Sci. Lett., 105, 368 –
377, doi:10.1016/0012-821X(91)90178-K.
Sparks, D. W., E. M. Parmentier, and J. Phipps Morgan (1993), Threedimensional mantle convection beneath a segmented spreading center:
Implications for along-axis variations in crustal thickness and gravity,
J. Geophys. Res., 98, 21,977 – 21,995, doi:10.1029/93JB02397.
Tolstoy, M., A. Harding, and J. Orcutt (1993), Crustal thickness on the MidAtlantic Ridge: Bulls-eye gravity anomalies and focused accretion,
Science, 262, 726 – 729, doi:10.1126/science.262.5134.726.
Turcotte, D. L. (1982), Magma migration, Annu. Rev. Earth Planet. Sci., 10,
397 – 408, doi:10.1146/annurev.ea.10.050182.002145.
van Keken, P. E., et al. (2008), A community benchmark for subduction
zone modeling, Phys. Earth Planet. Inter., 171, 187 – 197.
Wendt, J. I., M. Regelous, Y. L. Niu, R. Hekinian, and K. D. Collerson
(1999), Geochemistry of lavas from the Garrett Transform Fault: Insights
B11102
into mantle heterogeneity beneath the eastern Pacific, Earth Planet. Sci.
Lett., 173, 271 – 284, doi:10.1016/S0012-821X(99)00236-8.
Workman, R. K., and S. R. Hart (2005), Major and trace element composition of the depleted MORB mantle (DMM), Earth Planet. Sci. Lett., 231,
53 – 72, doi:10.1016/j.epsl.2004.12.005.
Yang, H. J., R. J. Kinzler, and T. L. Grove (1996), Experiments and models
of anhydrous, basaltic olivine-plagioclase-augite saturated melts from
0.001 to 10 kbar, Contrib. Mineral. Petrol., 124, 1 – 18, doi:10.1007/
s004100050169.
M. D. Behn and J. Lin, Department of Geology and Geophysics, Woods
Hole Oceanographic Institution, Woods Hole, MA 02543, USA.
P. M. Gregg, Department of Geosciences, Oregon State University,
Corvallis, OR 97331, USA. (greggp@geo.oregonstate.edu)
T. L. Grove, Department of Earth, Atmospheric, and Planetary Sciences,
Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
16 of 16